JXB Advance Access originally published online on September 12, 2006
Journal of Experimental Botany 2007 58(4):827-838; doi:10.1093/jxb/erl115
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Field Applications for Stress Monitoring |
Use of thermal and visible imagery for estimating crop water status of irrigated grapevine*
1Institute of Soil, Water and Environmental Sciences, Agricultural Research Organization (ARO), The Volcani Center, PO Box 6, 50250 Bet Dagan, Israel
2Institute of Agricultural Engineering, Agricultural Research Organization (ARO), The Volcani Center, PO Box 6, 50250 Bet Dagan, Israel
3Crop Ecology Laboratory, Migal, PO Box 831, 11016 Kiryat Shmona, Israel
4Golan Research Institute PO Box 97, 12900 Katzrin, Israel
5The J Blaustein Institute for Desert Research, Ben-Gurion University of the Negev, Sede-Boker Campus 84990, Israel
To whom correspondence should be addressed. E-mail: marmoeller{at}yahoo.de
Received 1 May 2006; Accepted 5 July 2006
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Achieving high quality wine grapes depends on the ability to maintain mild to moderate levels of water stress in the crop during the growing season. This study investigates the use of thermal imaging for monitoring water stress. Experiments were conducted on a wine-grape (Vitis vinifera cv. Merlot) vineyard in northern Israel. Irrigation treatments included mild, moderate, and severe stress. Thermal and visible (RGB) images of the crop were taken on four days at midday with a FLIR thermal imaging system and a digital camera, respectively, both mounted on a truck-crane 15 m above the canopy. Aluminium crosses were used to match visible and thermal images in post-processing and an artificial wet surface was used to estimate the reference wet temperature (Twet). Monitored crop parameters included stem water potential (
stem), leaf conductance (gL), and leaf area index (LAI). Meteorological parameters were measured at 2 m height. CWSI was highly correlated with gL and moderately correlated with stem. The CWSI-gL relationship was very stable throughout the season, but for that of CWSI-stem both intercept and slope varied considerably. The latter presumably reflects the non-direct nature of the physiological relationship between CWSI and stem. The highest R2 for the CWSI to gL relationship, 0.91 (n=12), was obtained when CWSI was computed using temperatures from the centre of the canopy, Twet from the artificial wet surface, and reference dry temperature from air temperature plus 5 °C. Using Twet calculated from the inverted Penman–Monteith equation and estimated from an artificially wetted part of the canopy also yielded crop water-stress estimates highly correlated with gL (R2=0.89 and 0.82, respectively), while a crop water-stress index using ‘theoretical’ reference temperatures computed from climate data showed significant deviations in the late season. Parameter variability and robustness of the different CWSI estimates are discussed. Future research should aim at developing thermal imaging into an irrigation scheduling tool applicable to different crops. Key words: Canopy temperature, CWSI, energy balance, infrared thermography, irrigation scheduling, stem water potential, stomatal conductance, Vitis vinifera
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Water is a major factor affecting both quality and quantity of wine grapes (Vitis vinifera L.). Excessive water leads to increased vegetative growth and yield, but grape quality parameters, such as sugar content, pigment formation, acidity, and wood maturation of the wine are negatively affected (Van Leeuwen and Seguin, 1994; Schultz and Gruber, 2005). Severe water stress on the other hand will induce stomatal closure (Escalona et al., 1999), causing strongly reduced or no assimilative activity and shoot growth and negative effects on wood maturation and sugar content. Maintaining a slight-to-moderate water deficit and thus inducing a certain level of stress can be very beneficial in grapevine cultivation as it stimulates optimal quality parameters without significantly compromising yield. In order to achieve this, crop water status must be measured accurately and reliably with the aim of maintaining a predetermined level of mild stress. Crop water status is preferably estimated by the direct sensing of plant water-stress parameters, like stem water potential and leaf conductance, rather than soil-based measurement approaches, which are prone to cumulative errors, require many sensors and may not be representative due to soil heterogeneity (Jones, 1990, 2004; Schultz and Gruber, 2005). For most crops, with the possible exception of anisohydric species whose leaf water potential decreases at increased water stress (Tardieu and Simonneau, 1998), for example, cotton and sunflower, leaf conductance is very sensitive to actual crop water stress, and therefore may give a better indication of stress than tissue-based measures such as leaf and stem water potential (Jones, 2004). However, leaf conductance is sensitive to other factors, the spatial coverage of spot leaf conductance measurements made with porometers is very limited and leaf-to-leaf variations require much replication for reliable data, so porometry is unsuitable for commercial applications (Hsiao, 1990). As a consequence, despite its accuracy the application of porometry as a measure for crop water status has been limited to research studies.
From energy balance considerations, it can be shown that leaf temperature is inversely correlated with transpiration rate and stomatal conductance (Fuchs and Tanner, 1966; Fuchs, 1990; Jones, 1992). The usefulness of canopy temperature as a measure of ‘crop water stress’ was recognized in the 1960s (Tanner, 1963; Gates, 1964). The ‘Crop Water Stress Index’, CWSI (Idso et al., 1981; Jackson et al., 1981; Idso, 1982; Jackson, 1982), is based on the difference between canopy temperature, as measured by infrared thermometry (IRT), and that of a ‘non water-stressed baseline’ referring to the temperature of a well-watered crop. Despite robust results with the CWSI approach from arid and semi-arid regions, limitations of its use as a routine tool stem from its high sensitivity to climate factors such as radiation and wind speed (Jackson et al., 1988; Jones, 1999a), which are not included in the computation of CWSI, and from the need to establish crop-specific non-water-stressed baselines for different agroclimatic zones.
In order to overcome these difficulties, a normalized CWSI using natural wet and dry reference surfaces (Twet and Tdry, respectively), for example, wetted or fully transpiring section of the canopy, was proposed by Clawson et al. (1989). This approach was used by Jones et al. (1999b), and the successful use of wet and dry reference plants has been reported by Leinonen and Jones (2004). However, these natural surfaces might not necessarily be uniform and difficulties are likely to arise with regard to their reproducibility. Meron et al. (2003) addressed problems related to low uniformity and reproducibility by using a wet artificial reference surface (WARS) for estimating Twet and by taking the upper base line temperature, Tdry, as Tair+5 °C, based on earlier publications (Irmak et al., 2000). The successful use of these artificial references on a sub-plot scale has been reported by Cohen et al. (2005), who used thermal imagery for evaluating and mapping the leaf water potential of cotton under various irrigation regimes.
Another alternative measure of crop stress which can be derived from thermal imagery is the spatial variability of leaf temperature, which should be related to the spatial variability of gL. Fuchs (1990) derived the variance of leaf temperature from a theoretical analysis of plant energy balance, and showed that it should be directly related to stress. Jones (1999b; Jones et al., 2002) suggested that this measure could be appropriate for crops with uniform full cover.
Much of the early work on thermography and CWSI estimation was based on handheld thermometers, which measure a temperature average over a single target area. Soil, trunk or dead tissue might be included in the sample area, potentially leading to considerable errors in estimated canopy temperature, particularly in sparse vegetation (Moran et al., 1994; Inoue et al., 1994). Recent technological advances in thermal imagery offer the potential to acquire spatial information on surface temperature in high resolution. Thermal, in conjunction with visible and near infrared (NIR) images enable the exclusion of non-leaf material in the estimate of canopy temperature and the possibility of selecting specific parts of the canopy for water-stress estimation (Jones et al., 2004).
Despite these recent significant improvements in the hard- and software used in thermal imaging, there is a current lack of knowledge in linking remotely measured canopy temperature and CWSI to ‘true’ ground-based measures of crop water stress, such as leaf conductance and water potential in the leaf or stem. However, knowledge of these relationships is required in order to translate thermal imagery data accurately into water-stress estimates, which can then serve as irrigation decision support tools.
In the present study, the potential of using thermal and visible images for the in-field estimation of crop water status of grapevine under three different irrigation regimes was investigated. The specific aims were (i) to compare thermal based CWSI estimates with plant water status parameters, (ii) to test the performance of different reference surfaces for CWSI computation, and (iii) to develop models for estimating stem water potential and stomatal conductance based on thermal and visible images.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Field plots
The field experiments were carried out at Kibbutz Yiftah, Israel (33° 10' N; 35° 55' E; 475 masl) during the summer of 2005. Times refer to Israel daylight saving time. Measurements were made on grapevines (Vitis vinifera cv. Merlot) planted in 1996 in a N–S direction with a vine and row spacing of 1.5 m and 3.0 m, respectively. Crop height varied between 2.0 m and 2.5 m. Soil texture was 50% clay, 24% silt, and 26% sand, with a pH of 7.4, a lime content of 9.5% and an EC (saturated paste) of 0.8 dS m–1. Vines were irrigated using a drip system with one line per row (integral drippers, emitter spacing 0.5 m, flow rate 2.3 l h–1, pressure compensated, Netafim, Israel). Three different treatments were applied: mild water stress (I), moderate water stress (II), and severe water stress (III). In order to maintain these treatments along the season, stem water potential (
stem) was measured once or twice a week and irrigation was applied when this reached the following thresholds: –9.5 bar (I), –12.0 bar (II), and –14.0 bar (III). Both timing and amount of water applications were adjusted for each treatment individually on a trial and error basis in order to maintain stem water potential near these thresholds. Commencement of irrigation and total amount of applied water until harvest were as follows: 24 May, 454 mm (I), 1 July, 150 mm (II); and 26 July, 23 mm (III), which was 89%, 29%, and 5%, respectively, of seasonal crop water requirements (509 mm) of fully watered grapevines, computed after Allen et al. (1998) using a crop coefficient equal to 0.3, 0.7, and 0.45 in the initial, mid-season and late-season development stages, respectively. There were three parallel experimental rows of which the two outer rows served as guard rows while measurements were conducted in the central row. In all three rows, three replicates per treatment, each consisting of seven vines, were laid out randomly in three main blocks.
Image acquisition and processing
Acquisition of thermal and colour images:
Thermal images of the plots were taken with an uncooled infrared thermal camera between 11.30 h and 15.00 h on 23 June, 12 July, 25 July, and 9 August 2005. The camera (ThermaCAM model SC2000, FLIR systems) had a 320x240 pixel microbolometer sensor, sensitive in the spectral range of 7.5–13 µm, and a lens with an angular field of view of 24°. Digital colour images were acquired with a digital camera (DSC-F717, Sony Inc.) that was attached to the thermal camera. The two cameras were mounted on a truck-crane about 15 m above the canopy. The canopy height was about 2 m, so that the linear field of view at the canopy level was 2x(15–2)xtan(24°/2)=5.5 m (i.e. 1.7 cm pixel–1). This resolution enabled differences between leaves and soil, as well as the selection of pixels that contained sunlit leaves, to be distinguished. Aluminium crosses were placed in the camera's field of view in order to geo-reference the digital RGB and the thermal images (Fig. 1). A wet artificial reference surface (WARS) was also placed in the camera's field of view (Fig. 1). The WARS was constructed as follows (Meron et al., 2003). A 5 cm thick slab of expanded polystyrene foam was floated in a 40 cmx30 cmx12 cm plastic tray, covering most of the water surface. It was coated with a doubled piece of 0.5 mm thick water absorbent non woven polyester and viscose mixture cloth (Spuntech, Israel), overlaid on another 2 mm thick polyester non-woven water-absorbent cloth. The edges of the clothes served as a wick, soaking up water to replace evaporation, and the polystyrene foam insulated the float from the background. This floating set-up provided horizontal and vertical alignment and a permanently wet surface of reproducible radiometric and physical properties. The outer layer sensed by the IR image is thus insulated from the water bath and essentially an independent 
2 mm thick white water layer.
|
Processing of thermal images:
Thermal images were processed with digital image processing tools. The raw thermal images were obtained in the FLIR Systems' proprietary format and converted to grey-scale images as described by Cohen et al. (2005). The centres of the aluminium crosses were selected as geographical control points (GCPs) and the thermal and colour digital images were aligned and registered using Matlab R13 software (Mathworks Inc.). The colour image was used to select canopy pixels with specific features, such as sunlit pixels. This was performed by transforming the colour image from the RGB to the Hue-Saturation-Intensity (HSI) colour space and then by applying threshold values in each of the colour components. The above procedure yielded to a binary image where pixels belonging to the selected canopy fraction are represented by logical ‘one’ and all other pixels are represented by logical ‘zero’. The HSI threshold values were determined interactively by an operator while visually inspecting the resulting image. Once a set of threshold values was chosen, the consequent images were processed using the same set of threshold values. Using the above procedure two types of masks were created: masks of soil and masks of both soil and shadowed leaves (Fig. 2). A proprietary software was developed by our research group that implemented the above process, constructed the masks, and then used only pixels with a corresponding mask of logical ‘one’ in the statistical analysis of the temperature in the thermal images.
|
Meteorological and physiological measurements
Global radiation, wind speed, air temperature, and relative humidity were measured 2 m above-ground by a meteorological station positioned within the experimental plot. The sampling rate was 0.1 Hz and 1 min averages were logged to a data acquisition system (CR10X, Campbell Sci., Logan, Utah). In addition, hourly climate data and 24 h Penman–Monteith reference evapotranspiration are available from the Kadesh meteorological station, located less than 1 km from the experimental site.
Once or twice weekly, stem water potential stem was determined as follows: six shaded leaves per treatment in the lower part of the canopy were covered with aluminium foil-coated plastic bags at around 11.00 h for at least 90 min, in order to allow stem and leaf water potential to equilibrate (Naor, 1998). After this equilibration period, the leaves were removed and covered with a plastic bag and stem was measured using a pressure chamber (ARIMAD, Kfar Haruv, Israel), following the same procedure as outlined by Meron et al. (1987) for leaf water potential measurements in cotton.
Leaf area index LAI was estimated by Gap Fraction Inversion (Cohen et al., 1997, 2000) on 19 July and 8 August. Measurements were done at three different zenith angles on three vines per treatment using a linear photosensor array probe (model Sunlink, Decagon Inc., Pullman, WA).
On days on which thermal and RGB images were taken, plant physiological measurements were carried out as follows: immediately following thermal and RGB image acquisition in each treatment plot, stomatal conductance and stem water potential were measured in the two vines in the centre of the respective plot. stem was measured with four replications per treatment plot on 23 June and 25 July, and two replications on 12 July and 9 August, respectively. Stomatal conductance was measured in sun-exposed fully developed leaves in the upper part of the canopy using a steady-state porometer (Li-Cor model 1600, Lincoln, Nebraska, USA). There were 10 replications per treatment plot on 12 and 25 July and 15 and 20 replications on 9 August and 23 June 2005, respectively.
Theory and equations
Crop water stress index CWSI was calculated after Jones (1992) as:
 | (1) |
 | (2) |
is the psychrometric constant, Rni is net radiation, 
a is the density of air, cp is the specific heat of air, and 
is the slope of the saturation vapour-pressure curve.
Tdry was estimated by adding 5 °C to the measured dry bulb temperature, as suggested by Irmak et al. (2000) and used by Cohen et al. (2005). As a second estimate for Tdry, the concept of isothermal radiation was used (Jones, 1999a):
 | (3) |
Soil water deficit SWDi on day i was calculated from irrigation records Inet and estimated crop water use ETc:
 | (4) |
 | (5) |
FC and permanent wilting point WP at the site are estimated as 0.48 and 0.32, respectively (Achtnich, 1980), the rooting depth Drz is 1.5 m and p is the maximum allowable depletion of total available soil water that does not cause a reduction in evapotranspiration (=0.45 for grapevine; Allen et al., 1998). |
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Climate parameters and direct measurements of crop water status
An overview of the meteorological conditions prevailing during the period of thermal imaging and CWSI computation is given in Fig. 3, and the deviation of daily climate parameters on the four thermal imaging days from the seasonal average is presented in Table 1. Radiation measurements indicate clear sky conditions on all four days, except for the period from 11.25 h to 12.25 h on 25 July (Fig. 3a), when clouds modulated solar radiation. Temperatures were above the seasonal average on all four days, and 12 July was one of the hottest and driest days of the summer. During the period of the day when crop water stress measurements were made, both air temperature and vapour pressure deficit were relatively constant on 12 July and 9 August. By contrast, a steady 4 °C increase in air temperature was recorded on 23 June between 11.30 h and 14.10 h and a 3.5 K rise occurred on 25 July between 11.45 h and 14.20 h. Temperature increases on both days coincided with increases in vapour pressure deficit (VPD), which was particularly marked on 23 July, rising by 1.9 kPa from 11.35 h to 14.10 h. Wind speeds prevailing on these two days (23 June and 25 July) were relatively low, and increased slightly during the 4 h of thermal imaging. Daily wind speeds on 12 July and 9 August were higher than the seasonal average (Table 1), remaining relatively constant during the 4 h of measurements on 9 August, but rising sharply on 12 July at around 12.30 h, which indicates the onset of the Mediterranean sea-breeze on that day. These data represent a relatively wide range of climatic conditions for Israel's predominantly clear-sky summer.
|
|
The seasonal course of measured midday stem water potential is shown in Fig. 4.
stem was significantly higher (t test, P=0.05) in treatment I (crosses) than in the other two treatments (open triangles and filled circles) starting on 20 June, some 27 d after commencement of irrigation in this treatment. Water applications in treatment II started on 1 July and stem was significantly higher than in treatment III immediately thereafter, except for 11 July. The seasonal minimum of stem was –9.5, –11.8, and –16.5 bar in treatments I, II, and III, respectively.
|
Average gL and
stem measured during thermal and colour image acquisition are summarized in Table 2. The effects of the stress treatments on stem and gL were highly significant on all four dates (ANOVA). However, differences in gL and stem between treatments II and III were only significant on the 2nd and 3rd thermal imaging days, respectively. A strong positive correlation found between stem and gL (R=0.84; significant at P <0.01) on a seasonal basis was even higher when individual days were considered (Table 3). The slope of the regression curve decreased over the 4 d of measurement, indicating that for a given stomatal conductance, a lower stem water potential was observed as time progressed. This fact is corroborated by daily average values of stem and gL in treatments II and III (Table 2). It is noted that the intercept of the empirical stem–gL relationships was nearly constant, but significantly lower on the last day of measurement (Table 3). A relatively low correlation coefficient was found on 12 July. This was accompanied by a relatively large standard error of some stem data points. The number of replicates per block on that date was two leaves as compared to four leaves on 23 June and 25 July, the range of stem values was small, and differences between treatments II and III were not significant (see also Table 2a). Therefore, it is presumed that the low correlation coefficient is due to inaccuracies in stem measurement.
|
|
Leaf area index
Treatment I (mild water stress) had the highest LAI on both days of measurement (Table 4; 19 July and 8 August), but differences between treatments II and III were not significant. In the severe water stress treatment III, LAI did not change significantly during the 20 d between measurements while total irrigation was 10 mm. LAI in treatments I and II increased by approximately 0.2, while irrigation was 117 and 71 mm, respectively. These results confirm that the irrigation regimes produced three long-term levels of crop water status.
|
Relationships between CWSI and measured crop water status
For estimation of Tcanopy, different sections of the canopy were used: all canopy, sunlit canopy, centre of the canopy, and sunlit leaves from the centre of the canopy. The centre of the canopy yielded the best results (i.e. highest correlation coefficients for the relationship between CWSI and gL) and there was no significant difference in correlation coefficients obtained from the centre of the canopy or its sunlit fraction. Therefore, in the results presented hereafter CWSI was based on temperature of the centre of the canopy.
Relationships between CWSI and stem water potential and between CWSI and leaf conductance are shown for individual days in Figs 5 and 6. High correlations were found between gL and stem and CWSI using Tdry from air temperature +5 °C and Twet from the wet artificial reference surface WARS. The slope of the CWSI–stem relationship varied over time with steeper slopes observed later in the season. Lower coefficients of determination were observed on 12 July and 9 August when the number of replications per plot was lower. On the other hand, both slope and intercept of the CWSI-gL relationship were found to be very stable over time (Fig. 6). A low coefficient of determination was found for the last day of measurement (9 August). It is noted that on that day leaf senescence, characteristic of the end of the growing season, was observed in some plots, and the hardy leaves selected for gL may not have been representative of the whole canopy measured on the images. The seasonal CWSI-gL relationship based on daily data per treatment is presented in Fig. 7. CWSI was highly correlated with measured stomatal conductance (R2=0.91, significant at P <0.0001).
|
|
|
Selection of different reference temperatures
The seasonal relationships between crop water stress indices based on different reference temperatures and measured stomatal conductance (Fig. 8) were all highly significant (P <0.01). When the last day of measurement (9 August) was not included in the analysis, the R2 fit of the model generally increased. As mentioned earlier, on 9 August many leaves in all treatments had already started to lose pigments and become dry, and therefore leaf conductance of the whole canopy was probably lower than that of the leaves selected for measurement with the porometer. This reasoning is supported by the fact that all data points for 9 August lie to the right of the regression line (except for CWSIwet_canopy in Fig. 8c), indicating that for a given measured leaf conductance, higher canopy temperature and hence water stress index was recorded on that day. Excluding these problematic data points had the least effect on the slope and intercept of the CWSI–gL relationship (Fig. 8a) and a relatively small impact on the slope of the CWSIT_wet–gL relationship (Fig. 8b) and the CWSIwet_canopy–gL relationship (Fig. 8c). Slope, intercept and quality (R2) of the relationship between the theoretical crop water stress index I2 and gL were significantly affected when the 9 August data were excluded but the CWSIwet_canopy was least affected by the late season points, both in terms of regression line fit (R2) and deviation of 9 August data from the other measurements. This is presumably due to the fact that unlike the other three indices, CWSIwet_canopy refers to a reference temperature extracted from the canopy image itself, thus incorporating seasonal effects on leaf colour and temperature. It can be shown mathematically that an equally increased wet reference temperature, induced by leaf colour changes, cancels out higher canopy temperatures, thus ‘shifting’ data points back to the left onto the regression line, as observed in the difference between Fig. 8d and c.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Treatment and climate differentiation
It is important to base the relationships between remotely determined CWSI (from thermal imaging) and directly measured crop water stress on a broad range of climatic and crop physiological conditions. This improves the quality of the established relationship and provides insight as to whether the model is robust at various crop growth stages. In this context, the CWSI–
stem relationship proved inferior to CWSI–gL (Figs 5, 7) because the former changed during the season. The difference between the two relationships highlights that physically CWSI is more closely related to gL than to stem. The shift in the CWSI–stem relationship may be from an adjustment in the plant response to water potential during the summer, which leads to higher leaf conductance for the same stem; perhaps due to differences in osmotic adjustment in the irrigation treatments. This seasonally changing stem–gL relationship was also reported for peach (Marsal and Girona, 1997) and pear trees (Marsal et al., 2002). Midday stem is commonly used for irrigation scheduling in commercial deciduous orchards in Israel (Naor, 2006), and was shown here to be highly correlated with SWD (Fig. 9), a parameter that is critical for irrigation management. Thus, the change in the CWSI–stem relationship during the season might require variable irrigation thresholds if CWSI is used for irrigation scheduling.
|
Full differentiation between the three water stress treatments occurred earlier in the season when leaf conductance is considered, as compared to stem water potential (Table 2). The influence of the irrigation treatments on the crop was also seen in LAI measured on two dates (Table 4). Measured
stem was thus less sensitive to the level of water stress than gL during the first half of the growing season, although it should be noted that stem estimation was based on a much smaller sampling size than were leaf conductance measurements. Covering a wide range of climatic conditions during the experimental stage is desirable to (i) establish that the model is accurate and stable under all conditions and hence (ii) increase model reliability and accuracy when applied to other sites. Given the relatively uniform clear sky and hot Israeli summer climate, in this experiment a relatively wide range of climate conditions was covered, ranging from very hot, dry and windy conditions (12 July) to moderate temperatures, humid and windy conditions (9 August). The radiation regime was only representative of clear sky conditions. It is therefore believed that the conditions were representative of summer climate in clear-sky vine-growing regions of the temperate and subtropical zones.
Variability of canopy temperature and conductance
Fuchs (1990) suggested, based on a theoretical analysis, that variability of canopy temperature increases with crop stress and that this relationship might be used as an indicator of stress. Some variability in this study's results might result from inclusion of hot background soil in pixels at the edge of the canopy. In order to overcome this, threshold temperatures were defined and only the central part of the canopy was analysed. The standard deviation of canopy temperature in the thermal images ranged from 1.6 °C to 3.8 °C per treatment and day or 5–13% of the respective mean. However, the relationship between canopy temperature variability and water stress (measured as leaf conductance) was weak and statistically insignificant (R2 <0.10; n.s.; data not shown). This agrees with conclusions drawn by Jones et al. (2002) that the Fuchs method (1990) might not be appropriate for row crops, such as grape vines, but rather might apply to homogeneous crops. However, an increased variability (expressed in the coefficient of variation) of stomatal conductance was observed at increased crop water stress (Fig. 10). It is not fully understood why this increased crop water stress was not reflected in increased variability of leaf temperature. Changes in leaf angle (away from the sun) and curling in stressed canopies might have prevented higher leaf temperatures, resulting in relatively uniform temperatures despite increased water stress and variability in leaf conductance.
|
Forecast quality and robustness of different CWSI estimates
Figure 8 demonstrates that CWSI, computed using Tair+5 °C and WARS for Tdry and Twet, respectively, was the most precise crop water stress estimate. In addition to its accuracy, the use of an easily reproducible wet reference surface, and no climate data requirements other than air temperature, potentially make it an attractive stress detection tool for precision agriculture. When Tdry and Twet were estimated from the inverted Penman–Monteith equation (equations 2, 3) solved with 1 min climate data (Fig. 8b, d) the R2 was lower. The use of Twet_canopy (Fig. 8c) as an estimate for Twet should be considered for crop varieties with seasonably strongly variable reflectance and pigment content, since CWSIwet_canopy proved to be the least affected by seasonal changes, although its overall accuracy was somewhat lower than that of CWSI (Fig. 8a). The quality of the seasonal CWSI–gL relationships was presumably negatively affected by late season changes in pigment content. If the linear models from these relationships were used for irrigation scheduling, they might lead to over-irrigation in the late season.
It can be deduced from Fig. 8 that for a measured CWSI, the model underestimates ‘true’ leaf conductance on 9 August, and overestimates water stress. This late season discrepancy was particularly marked in the theoretical crop water stress index I2, and might be of concern when applying CWSI for irrigation management in wine production where controlled stress is important for grape quality (Van Leeuwen and Seguin, 1994; Schultz and Gruber, 2005).
|
Conclusions |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
This study demonstrated that (i) fusion of thermal and visible imaging can improve the accuracy of remote CWSI determination and provide precise data on crop water status and stomatal conductance of grape vines; (ii) throughout the season CWSI computed with air temperature +5 °C and the artificial wet reference surface for Tdry and Twet, respectively, was the most robust stress index; while (iii) stable linear relationships existed also when Twet was derived from the energy balance (CWSIT_wet) and from image analysis of a wetted part of the canopy (CWSIwet_canopy).
Future studies should aim at (i) verifying and/or modifying the water stress models presented in this paper for other grape varieties and (ii) developing the CWSI–gL relationships into a usable tool for irrigation management (i.e. timing and amount of water application) on a field scale.
|
Acknowledgements |
|---|
This work was supported by grant No. TB-8006-04 from the Binational Agricultural Research and Development Fund and the Chief Scientist of the Israeli Ministry of Agriculture through project No. 458-0361-05.
|
Footnotes |
|---|
* Contribution No. 610/06 from the Agricultural Research Organization, The Volcani Center, Bet Dagan, Israel.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Achtnich W. (1980) BewässerungslandbauStuttgart Eugen Ulmer Verlag pp. 126–146.
Allen RG, Pereira LS, Raes D, Smith M. (1998) Crop evapotranspiration. Guidelines for computing crop water requirementsRome FAO Irrigation and Drainage Paper 56.
Clawson KL, Jackson RD, Pinter PJ. (1989) Evaluating plant water stress with canopy temperature differences. Agronomy Journal 81 858–863.
Cohen S, Rao RS, Cohen Y. (1997) Canopy transmittance inversion using a line quantum probe for a row crop. Agricultural and Forest Meteorology 86 225–234.[CrossRef][Web of Science]
Cohen S, Striem MJ, Bruner M, Klein I. (2000) Grapevine leaf area index evaluation by gap fraction inversion. Acta Horticulturae 537 87–92.
Cohen Y, Alchanatis V, Meron M, Saranga S, Tsipris J. (2005) Estimation of leaf water potential by thermal imagery and spatial analysis. Journal of Experimental Botany 56 1843–1852.
Escalona JM, Flexas J, Medrano H. (1999) Stomatal and non-stomatal limitations of photosynthesis under water stress in field-grown grapevines. Australian Journal of Plant Physiology 26 421–433.[Web of Science]
Fuchs M. (1990) Infrared measurement of canopy temperature and detection of plant water stress. Theoretical and Applied Climatology 42 253–261.
Fuchs M and Tanner CB. (1966) Infrared thermometry of vegetation. Agronomy Journal 58 597–601.
Gates DM. (1964) Leaf temperature and transpiration. Agronomy Journal 56 273–277.
Hsiao TC. (1990) Measurements of plant water status. In Stewart BA and Nielsen DR (Eds.). Irrigation of agricultural cropsMadison, WI American Society of Agronomy Inc. pp. 243–279.
Idso SB, Jackson RD, Pinter PJ Jr, Reginato RJ, Hatfield JL. (1981) Normalizing the stress-degree-day parameter for environmental variability. Agricultural Meteorology 24 45–55.
Idso SB. (1982) Non-water-stressed baselines: a key to measuring and interpreting plant water stress. Agricultural Meteorology 27 59–70.[CrossRef]
Inoue Y, Sakuratani T, Shibayama M, Morinaga S. (1994) Remote and real-time sensing of canopy transpiration and conductance: comparison of remote and stem flow gauge methods in soybean canopies as affected by soil water status. Japanese Journal of Crop Science 63 664–670.
Irmak S, Haman DZ, Bastug R. (2000) Determination of crop water stress index for irrigation timing and yield estimation of corn. Agronomy Journal 92 1221–1227.
Jackson RD. (1982) Canopy temperature and crop water stress. Advances in Irrigation Research 1 43–85.
Jackson RD, Idso SB, Reginato RJ, Pinter PJ Jr. (1981) Canopy temperature as a drought stress indicator. Water Resources Research 17 1133–1138.
Jackson RD, Kustas WP, Choudhury BJ. (1988) A re-examination of the crop water stress index. Irrigation Science 9 309–317.[CrossRef]
JMP BJ. (2002) Version 5.0.1Cary, NC SAS Institute Inc. pp. 1989–2002.
Jones HG. (1990) Plant water relations and implications for irrigation scheduling. Acta Horticulturae 278 67–76.
Jones HG. (1992) Plants and microclimate 2nd edn Cambridge Cambridge University Press.
Jones HG. (1999a) Use of infrared thermometry for estimation of stomatal conductance as a possible aid to irrigation scheduling. Agricultural and Forest Meteorology 95 139–149.[CrossRef]
Jones HG. (1999b) Use of thermography for quantitative studies of spatial and temporal variation of stomatal conductance over leaf surfaces. Plant, Cell and Environment 22 1043–1055.[CrossRef]
Jones HG. (2004) Irrigation scheduling: advantages and pitfalls of plant-based methods. Journal of Experimental Botany 55 2427–2436.
Jones HG, Stoll M, Santos T, de Sousa C, Chaves MM, Grant OM. (2002) Use of infrared thermometry for monitoring stomatal closure in the field: application to grapevine. Journal of Experimental Botany 53 2240–2260.
Leinonen I and Jones HG. (2004) Combining thermal and visible imagery for estimating canopy temperature and identifying plant stress. Journal of Experimental Botany 55 1423–1431.
Marsal J and Girona J. (1997) Relationship between leaf water potential and gas exchange activity at different phonological stages and fruit load in peach trees. Journal of the American Society for Horticultural Science 3 415–421.
Marsal J, Mata M, Arbonés A, Rufat J, Girona J. (2002) Water stress limits for vegetative and reproductive growth of ‘Bartlett’ pears. Acta Horticulturae 596 659–664.
Meron M, Tsipris J, Charitt D. (1987) Pressure chamber procedures for leaf water potential measurements of cotton. Irrigation Science 8 215–222.
Meron M, Tsipris J, Charitt D. (2003) Remote mapping of crop water status to assess spatial variability of crop stress. In Stafford J and Werner A (Eds.). Precision agriculture. Proceedings of the 4th European Conference on Precision Agriculture, Berlin, GermanyWageningen Academic Publishers pp. 405–410.
Moran MS, Clarke TR, Inoue Y, Vidal A. (1994) Estimating crop water deficit using the relation between surface-air temperature and spectral vegetation index. Remote Sensing of Environment 49 246–263.[CrossRef]
Naor A. (1998) Relationships between leaf and stem water potentials and stomatal conductance in three field-grown woody species. The Journal of Horticultural Science and Biotechnology 73 431–436.
Naor A. (2006) Irrigation scheduling and evaluation of tree water status in deciduous orchards. Horticulture Reviews 32 111–165.
Schultz HR and Gruber BR. (2005) Bewässerung und ‘Terroir’. Ergänzung oder Gegensatz? Das Deutsche Weinmagazin 1 24–28.
Tanner CB. (1963) Plant temperatures. Agronomy Journal 55 210–211.
Tardieu F and Simonneau T. (1998) Variability among species of stomatal control under fluctuating soil water status and evaporative demand: modeling isohydric and anisohydric behaviours. Journal of Experimental Botany 49 419–432.[Abstract]
Van Leeuwen C and Seguin G. (1994) Incidences de l'alimentation en eau de la vigne, appreciée par l'etat hydrique du feuillage, sur le développement de l'appareil végétatif et la maturation du raisin (Vitis vinifera variété Cabernet franc, Saint-Emilion 1990). Journal International Sciences de la Vigne et du Vin 28 81–110.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. Vincent, A. Ergul, M. C. Bohlman, E. A. R. Tattersall, R. L. Tillett, M. D. Wheatley, R. Woolsey, D. R. Quilici, J. Joets, K. Schlauch, et al. Proteomic analysis reveals differences between Vitis vinifera L. cv. Chardonnay and cv. Cabernet Sauvignon and their responses to water deficit and salinity J. Exp. Bot., May 1, 2007; 58(7): 1873 - 1892. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||